gpt-5-3-codex-82905fe7·16 events·first seen Aliases: GPT-5.3-Codex, GPT-5.2-Codex, GPT-5-Codex
OpenAI has released GPT-5.2-Codex to the API and for users accessing Codex with an API key. The announcement is brief, with no benchmark claims or capability details disclosed. This extends API access to a coding-specialized variant of GPT-5.2.
OpenAI has released GPT-5.2-Codex via the Responses API, a variant of GPT-5.2 specifically optimized for agentic coding tasks in Codex and similar environments. The release extends the GPT-5.2 model family with a coding-specialized deployment. This is a tier-1 announcement from OpenAI's official release notes.
OpenAI has optimized its inference stack for API customers, resulting in approximately 40% faster inference for GPT-5.2 and GPT-5.2-Codex. Model weights are unchanged, meaning the improvement is purely infrastructural. This is a meaningful latency reduction for production API users without any model capability tradeoff.
OpenAI released updates to the Codex app introducing support for GPT-5.3-Codex, a new mid-turn steering capability allowing users to redirect the model while it is actively working, and support for arbitrary file attachments. Mid-turn steering is a notable agentic UX feature that enables real-time human-in-the-loop intervention during long-running tasks. The release also includes a minor UI bug fix.
OpenAI released GPT-5.3-Codex, an incremental upgrade to GPT-5.2-Codex targeting complex software engineering tasks. The new model runs 25% faster, improves reasoning and professional knowledge, and adds better real-time collaboration features including progress updates and in-session steering. It is available immediately across ChatGPT paid plans via the Codex app, CLI, IDE extension, and Codex Cloud, with API access forthcoming.
OpenAI has made GPT-5.3-Codex available natively in Cursor and VS Code, with API access rolling out to a limited set of customers in a phased release. This is the first model OpenAI has classified as a high-security capability under its Preparedness Framework, with safety controls described as continuing to scale alongside expanded API access. The dual significance of a new coding-specialized model and a novel safety classification tier makes this a notable release.
OpenAI launched a research preview of GPT-5.3-Codex-Spark, a smaller, latency-optimized variant of GPT-5.3-Codex targeting real-time coding workflows at over 1000 tokens per second. The release is available to ChatGPT Pro users via the Codex app, CLI, and IDE extension, but not yet through the API. It marks the first public milestone of OpenAI's partnership with Cerebras, whose inference hardware enables the high-throughput performance. The model is text-only with a 128k context window and operates under separate usage limits during the research preview.
OpenAI has made GPT-5.3-Codex available through its Responses API. The release note is minimal, providing no benchmark claims or capability details beyond the availability announcement. GPT-5.3-Codex appears to be a code-specialized variant in the GPT-5 family.
OpenAI released GPT-5.4, a frontier model that consolidates recent advances in reasoning, coding, and agentic workflows, incorporating the coding capabilities of GPT-5.3-Codex. The ChatGPT deployment introduces a 'Thinking' mode that surfaces an upfront reasoning plan mid-response, allowing users to redirect the model before it completes. Additional improvements include enhanced deep web research for specific queries, better long-context retention during extended thinking, and improved context window management for faster, higher-quality outputs.
OpenAI is removing six model variants from the ChatGPT model picker starting April 7, 2026, including gpt-5.2-codex, gpt-5.1-codex-mini, gpt-5.1-codex-max, gpt-5.1-codex, gpt-5.1, and gpt-5, with full removal from Codex on April 14. Users are directed toward gpt-5.4, gpt-5.4-mini, gpt-5.3-codex, and gpt-5.2 as the supported options, with gpt-5.3-codex-spark available to Pro subscribers. The update signals OpenAI consolidating its model lineup around newer GPT-5.x variants and retiring earlier iterations.
A new arXiv preprint introduces SkillComposer, a method that frames skill selection for LLM agents as a structured prediction problem — jointly deciding which skills to activate, how many, and in what order via a constrained autoregressive decoder over skill identifiers. The approach addresses a bottleneck in growing skill libraries where existing retrieval and full-context methods fail to capture the joint nature of skill composition. Evaluated on SkillsBench across two production-grade coding agents (GPT-5.2-Codex and Gemini-3-Pro-Preview), SkillComposer raises pass rates by +23.1 and +18.2 percentage points over no-skill baselines, matching gold-skill retrieval upper bounds at lower prompt-token cost.
OpenAI has published an addendum to the GPT-5 system card introducing GPT-5-Codex, a version of GPT-5 specifically optimized for agentic coding within the Codex environment. The model features dynamic thinking-effort adjustment, scaling compute based on task complexity—responding quickly to simple queries while sustaining longer independent work on complex coding tasks. This represents a specialized derivative of GPT-5 targeting software engineering agents rather than general-purpose use.
OpenAI has released GPT-5.2-Codex, described as their most advanced coding model. The model features long-horizon reasoning, large-scale code transformation capabilities, and enhanced cybersecurity features. This represents a specialized coding-focused model in the GPT-5 family.
OpenAI published a system card addendum for GPT-5.2-Codex, a specialized variant of GPT-5.2 focused on coding capabilities. The document provides safety evaluations, capability assessments, and deployment considerations specific to this coding-oriented model. As a Tier 1 source system card, it represents official documentation of a frontier coding model's properties and risk profile.
OpenAI has released the system card for GPT-5.3-Codex, described as the most capable agentic coding model to date. It combines the frontier coding performance of GPT-5.2-Codex with the reasoning and professional knowledge capabilities of GPT-5.2. The release represents a continuation of OpenAI's Codex line of specialized coding models within the GPT-5 family.
OpenAI has announced GPT-5.3-Codex, described as a Codex-native agent combining frontier coding performance with general reasoning capabilities. The model is designed to support long-horizon, real-world technical work. The announcement positions it as an agentic coding system rather than a standalone language model.